错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Deep Medical Imaging, Analysis the Brest Cancer Mammography

  • Wajih Rhalem,
  • Sory Millimono,
  • Larbi Bellarbi,
  • Sarah Maghzaz,
  • Najib Alidrissi,
  • Salsabil Hamdi,
  • Nassim Kharmoum,
  • Ahmed Benahou,
  • Said Jidane,
  • Lahcen Belyamani,
  • Hassan Ghazal,
  • Mostafa Ezziyani

摘要

Breast cancer is the most commonly diagnosed cancer in women both before and after menopause. One in nine women will be diagnosed with breast cancer during their lifetime, and one in 27 women will die from it. This makes the diagnosis of this cancer a public health priority. Over the past 20 years, several diagnostic techniques have been proposed, including the method based on breast mammography, which is widely used today and allows the specialist doctor to make decisions based on the patient's scan images. To improve diagnostic outcomes, several artificial intelligence methods are used today, notably Convolutional Neural Networks, which have been particularly interesting in recent studies. In this work, we conducted a deep analysis on breast images from the “mias” database to classify people with malignant or benign cancer with great precision. We used pre-trained architectures (XCEPTION, VGG, RESNET, and INCEPTION) to experiment with transfer learning and achieved performance of up to 99.18% accuracy, 99% recall, and 100% precision.